Mg/Ti multilayers: Structural and hydrogen absorption properties

Mg-Ti alloys have uncommon optical and hydrogen absorbing properties, originating from a “spinodal-like” microstructure with a small degree of chemical short-range order in the atomic distribution. In the present study we artificially engineer short-range order by depositing Pd-capped Mg/Ti multilayers with different periodicities. Notwithstanding the large lattice mismatch between Mg and Ti, the as-deposited metallic multilayers show good structural coherence. On exposure to ${\text{H}}_2$ gas a two-step hydrogenation process occurs with the Ti layers forming the hydride before Mg. From in situ measurements of the bilayer thickness $\Lambda$ at different hydrogen pressures, we observe large out-of-plane expansions of Mg and Ti layers on hydrogenation, indicating strong plastic deformations in the films and a consequent shortening of the coherence length. On unloading at room temperature in air, hydrogen atoms remain trapped in the Ti layers due to kinetic constraints. Such loading/unloading sequence can be explained in terms of the different thermodynamic properties of hydrogen in Mg and Ti, as shown by diffusion calculations on a model multilayered systems. Absorption isotherms measured by hydrogenography can be interpreted as a result of the elastic clamping arising from strongly bonded Mg/Pd and broken Mg/Ti interfaces.

Figure 1

(Color online) Sample geometry: the total multilayers thickness is 60 nm and, in every sample, Mg layers are twice as thick as Ti ones: $N\times \left[\text{Ti}\left(\frac{20}{N}\text{nm}\right)\text{Mg}\left(\frac{40}{N}\text{nm}\right)\right]$ with $N=40$, 20, 10, 5, 2, and 1.

Figure 2

(Color online) Measured (dots) and fitted (lines) XRD patterns of $N\times \left[\text{Ti}\left(\frac{20}{N}\text{nm}\right)\text{Mg}\left(\frac{40}{N}\text{nm}\right)\right]$ multilayers with $N=40$, 20, 10, 5, 2, and 1, covered with 5 nm of Pd. The simulations are based on idealized step model, not capturing the influence of defects and other imperfections.

Figure 3

(Color online) (a) X-ray reflectivity pattern of the $10\times$ sample, $10\times \left[\text{Ti}\left(2\text{nm}\right)\text{Mg}\left(4\text{nm}\right)\right]\text{Pd}\left(10\text{nm}\right)$, taken at 333 K in vacuum (base $\text{pressure}={10}^{-7}\text{Pa}$). The solid line is a simulation of the multilayer using GENX (Ref. 22). The inset shows the XRD pattern measured at high angles on a $10\times$ sample deposited in the same run (dots) and the corresponding fit (line). (b) Real part of the SLD profile corresponding to the simulation in Fig. 3a.

Figure 4

(a) Cross-section bright-field STEM image of a $20\times \left[\text{Ti}\left(2\text{nm}\right)\text{Mg}\left(4\text{nm}\right)\right]$ multilayer deposited on a Si(100) substrate and covered with 10 nm of Pd. (b) Intensity profile of the area delimited by a white dashed line.

Figure 5

(a) Cross-section TEM image of the $20\times$ sample shown in Fig. 4. The dashed area highlights a region in which structural coherence is maintained across multiple Mg/Ti interfaces. (b) HRTEM image of sample $20\times$, taken approximately in the middle of the multilayer. The dashed arrow, parallel to the [001] direction of the hcp structure, indicates the growth direction of the multilayer. (c) Fourier transform of Fig. 5b.

Figure 6

XRD patterns measured during loading (left) and unloading (right) of the $10\times$ sample, $10\times \left[\text{Ti}\left(2\text{nm}\right)\text{Mg}\left(4\text{nm}\right)\right]\text{Pd}\left(10\text{nm}\right)$. The intensities are normalized by the measuring times: 1 h for the “as deposited” and “hydrogenated” states, 5 min otherwise. Circles: simulations of a perfect hcp-(002) Mg/fcc-(111) ${\text{TiH}}_2$ $10\times$ multilayer.

Figure 7

XRR from the partially loaded $10\times$ multilayer at $p_{{\text{H}}_2}=6\text{Pa}$ and $T=333\text{K}$. The inset shows the measured linear dependence of the ratio ${\text{sin}}^2{\theta }_n/{\lambda }_{\text{Cu}K{\alpha }_1}^2$ versus $n^2$.

Figure 8

(Color online) XRR pattern of the fully loaded $10\times$ sample at $p_{{\text{H}}_2}={10}^3\text{Pa}$ and $T=333\text{K}$. The solid red line is a simulation of the multilayer using GENX (Ref. 22). (b) SLD profile corresponding to the simulation in Fig. 8a.

Figure 9

(Color online) Rocking curves measured over the second-order reflectivity peak, for the as deposited, intermediate, and fully hydrogenated $10\times$ sample. The scans are offset and only 50% of the experimental points are shown for clarity.

Figure 10

XRD patterns measured for the $40\times$, $20\times$, $10\times$, $5\times$, $2\times$, and $1\times$ multilayers after hydrogenation in ${10}^5\text{Pa}$ at room temperature.

Figure 11

SAED patterns for the hydrogenated state of a $10\times$ Mg/Ti multilayer. (a) Short- and (b) long-beam exposures.

Figure 12

(Color online) Loading pressure-transmission-isotherms measured by hydrogenography at 333 K for the Mg/Ti multilayers covered with 10 nm of Pd.

Figure 13

(Color online) Simulated time evolution (in logarithmic scale) of the hydrogen concentration in each layer of a $3\times \left[\text{Ti}\left(z\right)\text{Mg}\left(2z\right)\right]\text{Pd}\left(z\right)$ sample, on hydrogen absorption $\left(e_{\text{H}}=0\right)$ and desorption $\left(e_{\text{H}}=-20\right)$. The vertical dashed lines represent the intermediate states, observed experimentally both on loading and unloading, in which only the Ti layers are hydrogenated. The white dashed-dotted lines are the simulated desorption rate from the Ti layers at 433 K.